Video surveillance systems are commonly used in many applications. In some cases, video surveillance systems incorporate object detection algorithms (e.g., facial detection algorithms), however, such algorithms frequently fail to detect some of the objects in a video scene. Furthermore, the computing resources of some surveillance systems may prohibit the use of improved algorithms that, although more accurate, also may require additional computing resources.